Validation of a Predictive Rule for Severe Acute Lower Intestinal Bleeding

Meeting: Hospital Medicine 2012, April 1-4, San Diego, Calif.

Abstract number: 97618


Acute lower gastrointestinal bleeding (LGIB) accounts for about 30 percent of episodes of gastrointestinal bleeding. It encompasses a wide spectrum of symptoms ranging from trivial hematochezia to massive hemorrhage with shock. Identifying those who will later develop severe bleeding is a challenge. Strate and colleagues have identified predictors of severity in patients with acute lower intestinal bleeding. The aim of this study was to provide external validation for the Strate clinical prediction rule for severe acute LGIB.


We retrospectively selected patients with a principal diagnosis of LGIB between January 2007 and August 2011 at a busy urban hospital. Predictor variables were those from Strate’s original study: heart rate = 100/min, systolic blood pressure <= 115 mmHg, syncope, nontender abdominal exam, rectal bleeding in the first 4 h of evaluation, aspirin use, and >2 comorbid conditions. Severe LGIB was defined as the presence of any of the following: transfusion of =2 units of red blood cells in the first 24 h, recurrent rectal bleeding after 24 h of stability (requiring additional blood transfusions), need for ICU placement or readmission for acute LGIB within one week of discharge. An overall risk score was created by adding the number of predictive factors present. Patients were stratified into three risk groups: low (no risk factors), moderate (1–3 risk factors), and high (>3 risk factors).


Analysis was performed on 378 patients. The risk of severe bleeding in each category was: low risk 31%, moderate risk 55%, and high risk 84%. The area under the receiver operating characteristic curve (AUROC) was 0.70 with good calibration (H–L Chi square = 0.98). The magnitude of the risk score was significantly correlated with need for blood transfusions, ICU placement and length of stay but not with need for surgery or death. Inclusion of additional variables including age, gender, actual heart rate and systolic blood pressure did not improve the AUROC for the prediction of severe bleeding.


Strate and colleagues developed a promising clinical prediction rule for acute severe LGIB. The risk factors are simple and available within 4 h of presentation. However, we find the predictive ability of the rule to be lower than in the original study as evidenced by the lower AUROC (0.70 vs 0.75). The Strate LGIB score does not have sufficient discriminatory power to improve the triage of patients to appropriate levels of care, nor promote a more standardized approach to acute LGIB. Further research is needed to improve our ability to identify patients presenting with LGIB who are at high risk for severe bleeding.

To cite this abstract:

Mattour A, Khanshour A, Salami A, Buran G, Parekh R. Validation of a Predictive Rule for Severe Acute Lower Intestinal Bleeding. Abstract published at Hospital Medicine 2012, April 1-4, San Diego, Calif. Abstract 97618. Journal of Hospital Medicine. 2012; 7 (suppl 2). Accessed May 26, 2019.

« Back to Hospital Medicine 2012, April 1-4, San Diego, Calif.